Verification of Floral Origin of Honey by SPME-GC-MS Analysis of Volatile Organic Compounds and Chemometrics

BARONI, M.V.; NORES, M.L.; DÍAZ, M.P.; CHIABRANDO, G.A.; COSTA, C.; WUNDERLIN, D.A.

Praga, (Rep. Checa)

Workshop; II° Meeting of TRACE Project EU.; 2006

Union Europea

The floral origin of a honey is one of the principal characteristics that determine its traceability and quality. Among different chemical markers used to evaluate floral origin of honey, the profile of Volatile Organic Compounds (VOCs) is relevant to both its authentication and quality. Considering the great number of VOCs that can be present in honey, its profile represents a fingerprint that can be used to determine the geographical and / or botanical origin of honey. The main goal of this work was to study the composition of VOCs in unifloral honey using solid phase micro extraction (SPME) and gas chromatography coupled to mass spectrometry (GCMS), determining the most relevant to its origin by chemometrics. We report the evaluation of floral origin of honey by analysis of its VOCs profile, joined to the use of combined pattern recognition techniques. Unifloral honey samples (n=42), from five different floral origins, were analyzed by head-space SPME, attached to GC-MS. Considering its occurrence, we selected 35 VOCs out of the entire chromatogram, constructing a database that was analyzed by hierarchical cluster analysis (HCA), stepwise discriminant analysis (SDA), and non parametric K nearest neighbor (KNN) as classification methods. Both HCA and SDA were used as an exploratory tool to point out those VOCs most important in discrimining honey origin. Thus, we selected 6 out of 35 VOCs initially considered. Afterward, we estimated the power of these six compounds to discriminate among different floral origins by KNN method, obtaining 93% right classification. VOCs were identified as: octanal, benzenacetaldehyde, 1-octanol, 2- methoxyphenol, nonanal and 2-H-benzopyran-2-one (coumarin) by comparison with both NIST MS-library and pure commercial standards. These compounds were quantified in honey samples of five different floral origins. Medicago sativa honey is characterized by high amounts of benzeneacetaldehyde, nonanal and 2-methoxyphenol; Prosopis spp is better represented by nonanal and octanal. Melilotus albus honey is distinguished by 2-H benzopyran-2-one, while Helianthus annuus honey shows high amounts of 2-methoxyphenol but low levels of other VOCs. Finally, Prosopis caldenia honey is characterized by 1-octanol. Results demonstrate that SPME coupled to GC-MS can be used satisfactorily for the analysis of VOCs profile in honey, which can be associated to chemometric techniques like HCA, SDA and KNN, providing with a feasible alternative to evaluate the botanical origin of honey. We obtained better results by studying VOCs profile instead of looking for a particular marker for each floral origin. This combine techniques could be useful in evaluating both geographical or specie origin in other foods presenting characteristic VOCs profiles.